Abstract: Printed Circuit Boards are mostly seen in all electronic products in daily life. PCBs are easily rejected due to different kinds of defects that slow down the manufacturing process. Defects like spurs, shorts, pinholes, and mouse bites cause issues like leakage of current and open circuits, which affects the performance of PCB. PCB manufacturers must give their utmost priority to delivering defect-free products to remain in the market among various competitors and to meet customer satisfaction. PCB analysts are employed in every industry to ensure that the defects are kept to a minimum or negligible level. But manual inspection of PCB defects is really very challenging, time-consuming, and an unreliable process. There are many defects to detect, which are extremely low tolerance for errors, and significant experience is required to reliably recognize and handle such flawed units. An automated solution for detecting and classifying defects in PCBs is in great demand. Automated defect detectors have significant roles in the quality control of PCB manufacturing sectors due to automated inspection adaptions instead of manual inspections. Defect recognition and individual component position detection are the major concerns in PCB inspection system design. Proper positioning of components improves the operating speed and system accuracy. In this article, a full PCB defect Detector used to automate the detection and classifying defects in printed circuit boards is discussed. It can be implemented in local PCB manufacturing industries. The proposed PCB inspection system functions with maximum accuracy in suitable illumination conditions. \textcopyright 2023 IEEE.

Abstract: The main objective of this work is to investigate the behavior of an Automated Optical Inspection (AOI) robustness algorithm employed in a high-yield Printed Circuit Board (PCB) assembly Flexible Manufacturing Cell (FMC) utilizing neural networks and the machine learning concept with a focus on the comparison between different modeling options and techniques to predict and actuate the occurrence of failures in electronic components at the run-time assembling in a high-yield production line. A locally Flexible Manufacturing Cell (FMC) or Flexible Manufacturing System (FMS) is an artificial intelligence (AI) composite structure where sensors and actuators (small devices compared to the involved AOI FMC system) interact with the hosting media, creating PCBs gaps imaging as a coupled effect of AOI algorithms and Mean Time Between Failures (MTBF) in the FCM structure. This effect typically creates imaging gaps at lower frequencies than those that would be observed in the PCBs. An introduction about colored PCBs electronic components background modeling, Kernel methods, multi-model algorithms, Kalman filters, and Mixture of Gaussians (MoG) is given. The main technique used regarding electronic component failure calculus is the Mahalanobis Distance, which is initially applied using classical image subtraction equations and a classical Bandlets board model. \textcopyright 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Abstract: In today\textquotesingle s electronics manufacturing, Printed Circuit Board (PCB) component welding faults are a critical issue that can significantly impact the reliability and functionality of devices. Accurate detection of anomalies in welded components not only improves the quality and efficiency of the manufacturing process but also helps reduce economic waste.This paper presents a data pipeline that combines classical machine learning and deep learning techniques with computer vision to identify potential soldering defects in PCB assemblies. The proposed method involves cleaning the data obtained from Automated Optical Inspection (AOI) systems to create a dataset suitable for machine learning tasks, such as classification and anomaly detection.We then use this cleaned data to develop two classification models: a Random Forest (RF) model and a Convolutional Neural Network (CNN). These models are applied to perform both binary classification-distinguishing between defective and non-defective components-and multi-class classification to identify specific types of defects. The models are also tested on noisy data to assess their reliability when exposed to data differing from the training set.The results show good accuracy across almost all tests, with the CNN model demonstrating greater resilience to noise in the test set. \textcopyright 2025 IEEE.

Abstract: Industry 4.0 is currently the goal of many factories, promoting manufacturing factories and sustainable operation. Automated Optical Inspection (AOI) is a part of automation. Products in the production line are usually inspected visually by operators. Due to human fatigue and inconsistent standards, product inspections still have defects. In this study, the sample component assembly printed circuit board (PCB), PCB provided by the company was tested for surface components. The types of defects on the surface of the PCB include missing parts, multiple parts, and wrong parts. At present, the company is still using visual inspection by operators, the PCB surface components are more complex. In order to reduce labor costs and save the development time required for different printed circuit boards. In the proposed method, we use digital image processing, positioning correction algorithm, and deep learning YOLO for identification, and use 450 images and 10500 components of the PCB samples. The result and contribution of this paper shows the total image recognition rate is 92\% and the total component recognition rate reaches 99\%, and they are effective. It could use on PCB for different light, different color backplanes, and different material numbers, and the detection compatibility reaches 98\%. \textcopyright 2023, ComSIS Consortium. All rights reserved.

Abstract: With the rapid advancement of automation and intelligence in the electronics manufacturing industry, the throughput of a single production line was grown exponentially. Although high efficiency brought significant cost and time advantages, it also led to two major challenges: (1) extremely low tolerance for error\textemdash any slight defect might have caused the entire product to be scrapped; (2) increasingly diverse and more concealed types of defects\textemdash bubble defects, internal chip defects, printed circuit board (PCB) defects, and specific process defects were continuously emerged, posing significant challenges to the inspection process. Traditional manual visual inspection or single-task deep learning models were often struggled to balance detection efficiency and accuracy in complex industrial scenarios. To address the above challenges, a single-stage industrial defect detection model based on multi-dataset mixed training\textemdash MSAN-Net\textemdash was proposed in this paper. Representative datasets covering the typical scenarios mentioned above were collected and organized, and part of the data was re-annotated to ensure a high level of consistency with actual production environments. MSAN-Net was adopted an integrated architecture, deeply combining UnifiedViT, C2f modules, convolution operations, SPPF structure, and Bi-Level Routing Attention mechanism to achieve accurate identification of complex industrial defects. Extensive experiments (including comparisons with multiple methods, ablation studies, and external validation) showed that MSAN-Net was outperformed existing SOTA models in industrial defect detection tasks, significantly improving detection accuracy for multi-class defects in complex scenarios, reducing reliance on manual inspection, and effectively lowering scrap losses caused by defects, thus providing a reliable solution for intelligent quality inspection in the electronics manufacturing industry. \textcopyright 2013 IEEE.

Abstract: In the last decades, machine vision and Machine Learning (ML) techniques have seen significant improvements in developing new algorithms thanks to the increment of hardware performance. Exploiting machine vision for specific technological applications became an essential opportunity to introduce significant improvements in the manufacturing context. This paper proposes a study to analyze the ML capabilities to perform Automated Optical Inspection (AOI) for quality control in the manufacturing of Printed Circuit Boards (PCBs). The study has been performed by testing Mask R-CNN and YOLOv8 algorithms and an open-source PCB dataset designed to evaluate other ML techniques. The chosen open-source dataset (i.e. PCB defect dataset released by Open Lab on Human\textendash Robot Interaction of Peking University, HRIPCB) individuates appropriate classes of products and related defects for the context of interest, resulting in a suitable dataset for the performance evaluation of tested algorithms. The challenge of this specific application is the recognition of the component boundaries that have submillimetric dimensions and are not clearly identifiable. The comparison between Mask R-CNN and YOLOv8 highlights that the Mask R-CNN performs better in defect detection (i.e., Missing Holes and Shorts). In particular, for the missing hole defects, for example, the mAP50-95 is 0.798 for Mask R-CNN and 0.261 for YOLOv8. Instead, for the short defects, mAP50-95 is 0.519 for Mask R-CNN and 0.399 for YOLOv8. This work has been carried out to gather know-how for further activity related to AOI for quality control in the PCB assembly employed in the aerospace field. \textcopyright The Author(s) 2025.

Abstract: Due to the rapid development of manufacturing capabilities and the general improvement in requirements for product quality, the role of quality inspection in the industrial production process is becoming increasingly important. Unlike the case for natural objects, detailed information is particularly crucial in defect classification and localization, resulting in poor performance of general object detectors on complex defect detection tasks. Therefore, this paper proposes a progressively refined redistribution pyramid network for visual defect detection in complex images, in which three novel components are designed. (1) The aligned dense feature pyramid network (AD-FPN) refines scale differences and performs efficient alignment, alleviating feature misalignment in FPN-based methods. (2) The phase-wise feature redistribution module (PFRM) enhances the interaction between features across layers, where global information is reassigned in a semantically adaptive manner. (3) The adaptive feature purification module (AFPM) helps the network distinguish defects from complex backgrounds, and its update is directly supervised by an auxiliary branch to accelerate convergence. These ideas are all implemented based on YOLOv5. Extensive experiments on the Tianchi fabric dataset, the publicly available surface defect dataset NEU-DET, and the PCB defect dataset show that our method outperforms other state-of-the-art defect detection methods on most evaluation metrics. In addition, experimental results on the remote sensing dataset RSOD and pothole image dataset also demonstrate the strong generalization ability of our method in other complex scenarios. \textcopyright 2022 Elsevier B.V.

Abstract: Technological advances in the production of printed circuit boards (PCBs) are increasing the number of components inserted on the surface. This has led the electronics industry to seek improvements in their inspection processes, often making it necessary to increase the level of automation on the production line. The use of machine vision for quality inspection within manufacturing processes has increasingly supported decision making in the approval or rejection of products outside of the established quality standards. This study proposes a hybrid smart-vision inspection system with a machine vision concept and vision sensor equipment to verify 24 components and eight screw threads. The goal of this study is to increase automated inspection reliability and reduce non-conformity rates in the manufacturing process on the assembly line of automotive products using machine vision. The system uses a camera to collect real-time images of the assembly fixtures, which are connected to a CMOS color vision sensor. The method is highly accurate in complex industry environments and exhibits specific feasibility and effectiveness. The results indicate high performance in the failure mode defined during this study, obtaining the best inspection performance through a strategy using Vision Builder for automated inspection. This approach reduced the action priority by improving the failure mode and effect analysis (FMEA) method. \textcopyright 2025 by the authors.

Abstract: Automated visual inspection on PCB boards is a critical process in electronic industries. Misalignment component detection is one of the challenging tasks in the PCB inspection process. Defects during the production process might include missing and misaligned components as well as poor solder connections. Inspection of PCB is therefore required to create practically defect-free products. There are various methods have been developed to perform this task in literature. The significance of this research is to propose an efficient with low-cost system is still require in small scale manufacturing to perform the misalignment or missing component detection on PCB boards. However, an efficient, low-cost system is still required in small-scale manufacturing to perform the misalignment or missing component detection on PCB boards. In this study, a real-time visual inspection system is developed for misalignment component detection. The proposed system consists of hardware and software frameworks. The hardware framework involves the setup of devices and modules. The software framework is composed of pre-processing and post-processing. In pre-processing, image enhancement is applied to remove noises from captured images and You Only Look Once (YOLO) object detector for components detection. Subsequently, the detected components are compared to the corresponding defined pattern using a template-matching algorithm. As experimental shown, the proposed system satisfies the requirement of missing component detection on PCB boards. \textcopyright 2023 - IOS Press. All rights reserved.

Abstract: Quality control is a paramount task for the manufacturing process at the current time. Nowadays, industrial companies must guarantee the quality of their products to keep competitiveness. This paper addresses the problem of controlling the quality of GSM chips during the soldering process of individual chips in PCB substrates, providing quality to the manufacturing stage. We propose an approach to detect solder mask faults in GSM chips, during the soldering stage. The proposed methodology is based on the use of Vision Transformers (ViT), which use self-attention mechanisms to capture the spatial relationships between image patches, allowing the model to learn hierarchical visual representations. Real and simulated experiments were carried out to validate the proposed approach. Results show the obtained accuracy of 95.83\%, using the proposed ViT-based inspection approach. Furthermore, the proposed approach presents high accuracy even regarding noisy and blurry images, resulting in an accuracy of 94.94\% and 94.17\% for Salt and Pepper and Gaussian noise, respectively, in the worst scenario. Experiments demonstrate reliability and robustness, optimizing the manufacturing. \textcopyright 2024 IEEE.

Abstract: The Printed Circuit Board (PCB) accommodates various Integrated Circuit (IC) components arranged in a specific layout of bond pads, lines, and tracks. Throughout the manufacturing process, irregularities or defects often occur during drilling, etching, stripping, and other stages, impacting the performance and functionality of the circuit board. Many of these defects are related to soldering pads and copper balance, identifying them through manual inspection is time-consuming and error-prone. This necessitates the use of Automated Optical Inspection (AOI). Practices like template matching often require two identical PCBs, which are compared using mathematical algorithms to detect differences. However, they are not resilient to viewpoint variations and non-rigid deformations. The current inspection process primarily focuses on rectifying PCB images captured with tilts ranging from 0\textdegree to 84\textdegree using homography principles. This correction process operates within a maximum run time of 7.96 s. The adjusted images then undergo analysis via a pattern-matching unit, where the system receives images of the same PCB pattern, each exhibiting different defects. Structural information mapping is performed using various spatial-domain feature-based matching algorithms. When evaluated using SSI and MSE metrics, the model achieved high matching percentages of 99.67\%, 99.75\%, and 99.30\%, and low error rates of 0.343\%, 0.358\%, and 0.721\% for three different types of PCB designs considered. Additionally, the model excels in precisely identifying the location of defects in the PCB images without using bounding boxes, in accordance with the description of the co-images through a segmentation approach. Overall, the proposed system effectively corrects skew, accurately detects abnormalities and outperforms traditional assessment systems. \textcopyright 2024 by the authors.

Abstract: The soldering process for aerospace applications follows stringent requirements and standards to ensure the reliability and safety of electronic connections in aerospace systems. For this reason, the quality control phase plays an important role to guarantee requirements compliance. This process often requires manual control since technicians’ knowledge is fundamental to obtain effective quality check results. In this context, the authors have developed a new open source dataset (SolDef\_AI) to implement an innovative methodology for printed circuit board (PCB) defect detection exploiting the Mask R-CNN algorithm. The presented open source dataset aims to overcome the challenges associated with the availability of datasets for model training in this specific research and electronics industrial field. The dataset is open source and available online. \textcopyright 2024 by the authors.

Abstract: The high memory storage capacity is one of the main characteristics of current electronic devices. In the manufacture of high storage memories, more complex operations are required in the semiconductor encapsulation process, making this production phase more susceptible to failures. The semiconductor encapsulation process is a sensitive operation, given the high cost of equipment and materials, in this sense, a rigorous inspection process is a key task to keep the productive activity viable. This work addresses the automatic fault detection in the silicon wafer soldering procedure, allowing the identification of faults before the epoxy resin molding process. In addition, it provides reasonable information about the soldering step, reducing waste and increasing the inspection volume. We propose a methodology to classify solder balls, in the soldering process of silicon wafers in Ball Grid Array (BGA), contained in substrates of Printed Circuit Board (PCB). This approach consists of two steps: i) Segmentation of solder balls from silicon wafer images; and ii) Classification of solder balls by applying a deep learning technique. The relationship between the visual features and the different welding conditions is learned by the proposed predictive model. The proposed approach was validated by carrying out real and simulated experiments, which obtained high precision results of 98.009\% when the Convolutional Neural Network (CNN) model was used. Furthermore, in the worst-case scenario, where Salt and Pepper and Gaussian noises were applied to the images, the proposed approach presented high accuracy of 97.803\% and 96.567\%, respectively. Based on the above and on the analysis of the obtained results in the experiments, this work demonstrated reliability and robustness, optimizing the manufacturing. \textcopyright 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Abstract: Quality inspection of solder connections in electronic circuit manufacturing is commonly performed using automatic optical inspection (AOI) technology. The utilization of deep learning in AOI has demonstrated high accuracy and fast computation, yet it requires expensive graphics processing unit (GPU) equipped computers. To enable broader and more cost-effective utilization, there is a need for an AOI model that can be embedded in simple central processing unit (CPU) based computers. To pursue this objective, several research efforts have been undertaken to develop AOI models based on classical machine learning techniques. However, the accuracy and speed achieved by these models have not yet matched deep learning-based AOI models. This study aims to enhance the computational efficiency of classic machine learning by processing only pixels containing textural information. The effectiveness enhancement is achieved through the application of log-polar transformation in the extraction of texture features using the gray level co-occurrence matrix (GLCM) to detect defects in trough hole technology (THT) solder joint connections. By transforming cartesian coordinates into polar coordinates, the textural areas to be detected assume a square shape, facilitating efficient texture feature extraction. To ensure a significant improvement in performance, a comparative performance evaluation is conducted on classic machine learning-based AOI models with and without the log-polar transformation. The texture features extracted from both models are classified using the support vector machine (SVM) method. Model testing and evaluation reveals that the proposed enhancement effort is capable of increasing accuracy levels to 95\% and reach computation time by 120 milliseconds. \textcopyright (2024), (Intelligent Network and Systems Society). All Rights Reserved.

Abstract: Advancement of technologies in the electronics industry has render decrease in electronic components sizes and increase in number of components on Printed Circuit Board (PCB). Industries specialize in manufacturing Printed Circuit Board Assembly (PCBA), also implementing manual visual inspection in In Process Quality Control (IPQC) verification process to ensure quality of products. Such technology advancement has increased workload of operators and time taken to perform inspection. This study is aimed to reduce time consumption and cognitive load of operators, while ensuring consistency of visual inspection during component verification process by utilizing deep learning models to perform object detection based automated optical inspection of images consisting electronic components. Three deep learning algorithms were used in the study, which are Faster R-CNN, YOLO v3 and SSD FPN. Both Faster R-CNN and SSD FPN utilized ResNet-50 backbone, whereas YOLO v3 was built with Darknet-53 backbone. Various input image dimension and image resizing options were explored to determine the best model for object detection. At the end of the study, SSD FPN with input image dimension resized to 640 \texttimes 640 by keeping image aspect ratio and with padding is concluded as the best localization and classification model to perform object detection for various types of components present in digital image. \textcopyright 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

Abstract: To design a project that can harness wiring faults in the small components and PCB of our daily usage components such as mobiles, laptops and other small devices. It implements the usage of Augmented Reality which is an emerging technology in the current research trend. AR technology superimposes digital information onto the physical world, providing real-time, interactive visualization. By integrating AR with advanced diagnostic algorithms, technicians can quickly and accurately identify wiring faults such as short circuits, open circuits, and insulation failures. The AR system uses a combination of cameras and sensors to capture the wiring layout and diagnose issues, overlaying visual indicators of faults directly onto the live feed of the wiring.Key features of the proposed system include real-time fault localization, guided troubleshooting procedures, and the ability to visualize complex wiring schematics overlaid on physical components. The use of AR minimizes the need for extensive training, reduces inspection times, and enhances the accuracy of fault detection. Additionally, the system can log detected faults and repair actions, providing valuable data for predictive maintenance and system optimization. \textcopyright 2024 IEEE.

Abstract: Segmentation networks based on deep learning are widely used in the field of industrial vision inspection, including for the precise segmentation of surface defects on printed circuit boards (PCBs). However, most previous studies have focused only on the utilization of defect samples with visible defects and underestimated the value of template samples without surface defects. In fact, template samples can provide sufficient prior information to identify defects and are not difficult to obtain in many manufacturing scenarios. Therefore, an adaptive feature reconstruction network (AFRNet) is proposed in this paper to utilize these two types of samples. Specifically, AFRNet consists of two main components: a Siamese encoder with shared parameters for extracting features from the input sample pair, and a symmetrical feature reconstruction module for adaptively fusing these extracted features. Similar image-level and feature-level fusion schemes, as well as spatial misalignment caused by unaligned sample pairs have been carefully studied. Extensive experiments on a real-world PCB surface-defect dataset confirm the effectiveness of the proposed method, demonstrating that it can significantly improve the segmentation performance of multiple baselines, such as DANet, PSPNet and DeepLabv3. \textcopyright 2022, The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature.

Abstract: In the rapidly evolving field of printed circuit board (PCB) manufacturing, automated optical inspection (AOI) systems play a critical role but often face challenges such as computational inefficiencies, high costs, and limited defect data. To address these issues, we propose an ensemble methodology that combines lightweight models with custom data augmentation techniques to enhance defect classification accuracy in real-time production environments. Our approach mitigates overfitting in small datasets by generating diverse models through advanced data augmentation and employing feature-specific validation strategies. These models are integrated into an ensemble framework, achieving complementary results that improve classification accuracy while reducing computational overhead. We validate the proposed method using two datasets: the general classification dataset CIFAR-10 and an on-site real-world PCB dataset. With our approach, the average accuracy on CIFAR-10 improved from 97.6\% to 98.2\%, and the accuracy on the PCB dataset increased from 81\% to 89\%. These results demonstrate the method’s effectiveness in addressing data scarcity and computational challenges in real-world manufacturing scenarios. By improving quality control and reducing waste, our method optimizes production processes and contributes to sustainability through cost savings and environmental benefits. The proposed methodology is versatile, scalable, and applicable to a range of defect classification tasks beyond PCB manufacturing, making it a robust solution for modern production systems. \textcopyright 2024 by the authors.

Abstract: Surface mount technology (SMT) is a method to mount components directly onto printed circuit boards (PCBs) and is widely used in low-cost and high-density electronic assemblies. Pick-and-place (P\&P) is a core procedure for component placing after the solder paste printing (SPP) process in SMT. Generally, the industry uses an automated optical inspection (AOI) machine to detect defects after the components are mounted. However, the AOI machine cannot discern the failures\textquotesingle root causes and offer reliable P\&P machine maintenance references. With the advent of Industry 4.0, machine learning (ML) methods can be applied to improve production line maintenance. Therefore, the traditional check-up process can be changed into a data-driven, predictive, and condition-based maintenance process. Production efficiency can be significantly increased. In this article, a multioutput classification-based defect diagnosis (MCDD) model has been developed to trace the root causes of defects by using the patterns discovered from the experiment data. The experiments with initial machine errors are conducted and investigation information is collected. Compared with the traditional root cause identification model, the developed model is easier to adjust and can achieve an overall classification accuracy of 84.5\%. Furthermore, the transfer learning method has been used to apply the trained model for one component to other components and can achieve an accuracy of 81.74\%. \textcopyright 2011-2012 IEEE.

Abstract: Printed circuit boards (PCBs) becoming more complex as technology advances, adding new components and changing their architecture. One of the most crucial quality control procedures is PCB surface inspection since even little flaws in a signal trace may have a significant detrimental effect on the system. It has always been difficult to determine the pass/fail criteria in traditional machine vision systems based on small failure samples, despite the advancements in sensor technology. Suggesting a sophisticated PCB inspection method built on a skip-connected convolutional auto encoder to address these issues suggested to enhance the PCB inspection system by using convolutional autoencoders. The original, fault-free photos and the damaged ones were used to train the deep autoencoder model. The defect location was then located by comparing the decoded images with the input image. Using proper image augmentation to enhance the model training performance in order to get over the tiny and uneven dataset in the early phases of production. Printed circuit boards, or PCBs, are essential parts of electronic gadgets and are very significant to the electronics sector. While ensuring PCB quality and reliability is crucial, manual inspection techniques are often labour and error-intensive. The proposed novel machine learning (ML)-based method for identifying PCB defects demonstrates a significant improvement in detection rates compared to traditional methods, offering a promising solution for the electronics manufacturing industry. \textcopyright 2024 IEEE.

Abstract: Automated Optical Inspection (AOI) plays a crucial role in printed circuit board (PCB) manufacturing for detecting defects during the production process. However, traditional AOI techniques face challenges in handling complex defects, particularly those involving image registration and template-dependent defects. This study aims to address these issues by proposing an optical flow-assisted registration and defect detection method. By leveraging optical flow technology, we effectively enhance the accuracy and reliability of image registration and defect detection. Experimental results demonstrate significant performance improvements of our approach on both the DeepPCB dataset and the real-world HDPCB dataset. \textcopyright 2024 IEEE.

Abstract: Industrial solutions for the visual inspection of printed circuit boards (PCBs) have been enhanced in recent decades to become part of standards of quality control in the manufacturing process. However, most of the methods used to detect defects in PCBs are ineffective for the detection of defects in electronic board assemblies whose components have been manually placed and soldered, causing a high false-positive rate. In this article, we present a methodology for implementing an automatic defect detection system for manually soldered components. The proposed system consists of three subsystems: image processing to highlight defects and features, automatic selection of regions of interests (ROIs), and classification based on a deep neural network trained under a specific catalogue of defects to search for. The details of the defect catalogue and the coding of the training database are presented. To validate the methodology, a case study was carried out in a local manufacturing company where an experimental platform with partially controlled lighting was built. The neural network was trained with an image database of multiple electronic cards. Although the percentage of truly defective board was low, as measured against all detected defective boards, this proves insignificant for the results since all defective boards in the test set were detected/rejected, hence constituting a 100\% detection ratio. \textcopyright 2024 Elsevier Inc. All rights reserved.

Abstract: Industrial defect detection is crucial for ensuring product quality and production efficiency, playing a pivotal role in advancing smart manufacturing. This paper reviews defect detection technologies for various industrial products, including metals, textiles, and printed circuit boards, and introduces an innovative classification system. It also offers a detailed analysis of recent developments and practical applications of large models in industry defect detection. First, the basic principles of industrial defect detection are outlined. The detection methods are then categorized into three main groups: traditional image processing, machine learning, and deep learning, with their principles, case studies, limitations, and future development directions analyzed. Traditional methods consist of image preprocessing, segmentation, and feature extraction. Machine learning methods are divided into point-distance-based, hyperplane-based, tree-based, and neural network-based classification algorithms. Deep learning models are classified into two types: accuracy-oriented and efficiency-oriented. The paper organizes industrial defect datasets by type (multi-product and single-product), evaluates data quality and availability, and summarizes common evaluation metrics for accuracy, efficiency by task requirements. It also compares the latest methods on two public datasets to guide further research in defect detection. Real-world examples illustrate the end-to-end process, from data processing and hardware configuration to model training and deployment, while exploring the value and limitations of these technologies from the perspective of industry stakeholders. Finally, a systematic analysis of the key challenges and corresponding solutions is presented at the data and performance levels, and looks forward to the future direction of technological development, highlighting innovative paths and application potentials. \textcopyright 2013 IEEE.

Abstract: Being in the competitive area of electronics manufacturing, PCB defects have to be detected for reliability of products and maintaining the quality standards. This paper deals with the new advanced methodology for PCB automatic inspection by using CNN-based methodology for the detection of defect on PCBs. The models were highly trained and tested with a huge dataset that was designed especially for the identification of defects in PCBs. For effective identification and classification of various defects like the irregularities, missing components, gaps and electrical shorts, this paper uses the CNN architecture. The achieved results of experiments with the model\textquotesingle s accuracy up to 95.57\%, precision up to 95.44\% and recall up to 93.76\% underscore the importance of achieving high performance. Ensuring such stable and accurate representation of the CNN - based method makes this approach valuable in improving the production workflow. This automatic defect detection system enables man inspection to be minimized. Hence the possibility of human error becomes low and thus forms a reliable solution for production efficiency improvement and strict quality control throughout the electronics manufacturing process by the manufacturers. \textcopyright 2024 IEEE.

Abstract: In the electronics manufacturing sector, Printed Circuit Boards (PCBs) are crucial components whose quality is directly linked to the performance and reliability of end products. With the ongoing trend towards higher density and miniaturization in electronic devices, the demands on PCB quality are increasingly stringent, making traditional manual inspection methods insufficient for meeting the needs of high efficiency and accuracy. In response to these challenges, this study introduces an innovative PCB defect detection approach utilizing deep learning, specifically the advanced DAB Deformable DETR model. This model, an evolution beyond the conventional DETR framework, integrates multi-scale convolution and anchor weight optimization. These enhancements not only elevate the model\textquotesingle s proficiency in identifying small and intricate defects but also streamline the training process, curtailing both the duration and resources necessitated. This method was rigorously tested and validated on a PCB defect dataset, and the experimental results indicate that it not only achieves a mean Average Precision (mAP) comparable to current advanced methods but also offers substantial advantages in terms of detection speed and model robustness. \textcopyright 2024 SPIE

Abstract: Defects in the printed circuit board assembly significantly impact product functionality and quality. Automated optical inspection (AOI) systems, employed by manufacturing quality control teams, are designed to accurately detect these defects in a timely manner, thereby reducing the underkill (false negatives) and overkill (false positives) rates. An AOI system requires optimal settings for resolution, brightness, camera angle, and data variety to ensure effective defect detection. However, consistently achieving these ideal conditions in a manufacturing environment presents challenges. Our proposed framework enhances defect detection through data preparation and detection modules, effectively addressing these manufacturing challenges. We developed one-and two-stage object detectors and assessed their performance using precision, recall, and intersection over union metrics. Our framework employs a diverse range of augmentation techniques to effectively train the defect detectors, enabling the expansion of a limited data set. The trained detectors are evaluated using real-world data. We assessed quality control plans across various confidence thresholds. At a 65\% confidence threshold, one-stage detector models did not exhibit any false negatives and had minimal false positives. The You Only Learn One Representation (YOLOR) model outperformed both one-stage and two-stage detectors, achieving 100\% precision and recall, a 96\% mIoU, and an impressive inference time of 11 ms, making it an ideal choice for high-production printed circuit board assembly lines. \textcopyright 2013 IEEE.

Abstract: Surface-mounted technology (SMT) welding defect detection plays a key role in the printed circuit board assembly (PCBA) production process, which affects the use of electronic products and cost. Previous works tend to realize defect detection with only defect samples and they assume that there are sufficient defect samples. However, defect samples are usually difficult to collect in real-life scenarios while enough template samples can be easily obtained. In addition, most existing works carry out defect detection based on benchmarks with simple backgrounds of PCBA, which is not suitable for PCBA with complex structures in modern electronic product manufacturing. To address the above issues, we propose a template-guidance Siamese network based on YOLO for SMT welding defect detection (TGSYOLO), which is deployed on a real SMT automatic optical inspection (AOI) system. First, the two-stream structure is introduced to extract deep features in defect images and template images, in which template features serve as guidance knowledge. Then, a template fusion Transformer (TFT) is proposed to model global features between detect and template features in the low-level stage, which could acquire long-range correlations to force the network to focus on potential defect regions. Next, to avoid the disappearance of tiny defect features during deep feature fusion, a multiscale attention feature pyramid network (MAFPN) is proposed to directly fuse defect semantic information from low-level features, which retains detailed expressions of defects through skip connection and obtains compact fusion features. Furthermore, we collect limited welding defect samples based on more complex PCBA backgrounds than previous works through a real SMT AOI system. Experiments on the limited dataset show that TGSYOLO could reach 0.985 of mAP@0.5, 0.885 of mAP@0.75, and 0.984 of F1, which is 0.008, 0.054, and 0.025 higher than other SOTA methods. Also, generalization experiments on the public DeepPCB show that TGSYOLO could still reach the best with 0.991 of mAP@0.5 and 0.89 of mAP@0.75, which proves that TGSYOLO has good generalization performance. \textcopyright 2011-2012 IEEE.

Abstract: Computer vision-empowered quality is an important feature in modern manufacturing. These systems can perform defect inspections faster and more accurately than humans. There has been a rise in adopting deep-learning-based solutions for visual inspection tasks. However, new data is still necessary for each specific problem, and precise dataset annotation can be time-consuming considering precision-demanding approaches such as segmentation or region detection. This investigation proposes a method inspired by class activation mapping-based (CAM) to enhance deep neural network-based image classification algorithms by outputting an attention map where the defective area is indicated. With additional lightweight operations, it helps to roughly localize defects place in the image without using pixel-wise annotations. The proposed method is tested on glass bottles (Oliena), printed circuit boards (PCB defects), and industrial machine tool component surface defects (BSData) datasets. \textcopyright 2023 IEEE.

Abstract: Defect detection is a critical component in ensuring the highest quality of printed circuit board manufacture. The work investigates the application of transfer learning to identify YOLOv8 design flaws. The YOLOv8 model performs exceptionally well, with mean Average Precision (mAP) of 98.3\%, Precision of 98.6\%, and Recall of 96.0\%. Numerous datasets provide high detection rates across a range of fault types following thorough testing and inference. This extensive evaluation demonstrates the robustness of the model by reliably identifying Fissure, Mousebite, EC-Short, MC-Spur, Breakout, Pinhole, and Copper defects. The study emphasizes how the YOLOv8 design can identify flaws, providing a dependable and effective way to guarantee the dependability and caliber of printed circuit boards throughout the production process. The foundation for utilizing YOLOv8 in industrial quality control is established by this study, which holds significant potential for improving defect detection and quality control in the manufacturing industry. \textcopyright 2023 IEEE.

Abstract: Recently, many companies have introduced automated defect detection methods for defect-free PCB manufacturing. In particular, deep learning-based image understanding methods are very widely used. In this study, we present an analysis of training deep learning models to perform PCB defect detection stably. To this end, we first summarize the characteristics of industrial images, such as PCB images. Then, the factors that can cause changes (contamination and quality degradation) to the image data in the industrial field are analyzed. Subsequently, we organize defect detection methods that can be applied according to the situation and purpose of PCB defect detection. In addition, we review the characteristics of each method in detail. Our experimental results demonstrated the impact of various degradation factors, such as defect detection methods, data quality, and image contamination. Based on our overview of PCB defect detection and experiment results, we present knowledge and guidelines for correct PCB defect detection. \textcopyright 2023 by the authors.

Abstract: An enhanced approach for detecting defects in Printed Circuit Boards (PCBs) using a significantly improved version of the YOLOv8 algorithm is proposed in this research, the proposed method is referred to as YOLOv8-TDD (You Only Look Once Version8-Targeted Defect Detection). This novel approach integrates cutting-edge components such as Swin Transformers, Dynamic Snake Convolution (DySnakeConv), and Biformer within the YOLOv8 architecture, aiming to address and overcome the limitations associated with traditional PCB inspection methods. The YOLOv8-TDD adaptation incorporates Swin Transformers to leverage hierarchical feature processing with shifted windows, enhancing the model’s efficiency and capability in capturing complex image details. Dynamic Snake Convolution is implemented to dynamically adapt filter responses based on the input feature maps, offering tailored feature extraction that is highly responsive to the varied textures and defects in PCBs. The Biformer, with bidirectional processing capability, enriches the model’s contextual understanding, providing a comprehensive analysis of the PCB images to pinpoint defects more accurately. Experimental results demonstrate that YOLOv8-TDD model, achieves a precision of 97.9\%, a mean Average Precision (mAP0.5) of 95.71\%. This enhanced model offers significant potential for practical applications in PCB manufacturing, promising to elevate quality control standards through more reliable defect detection. \textcopyright The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2024.

Abstract: This research proposes a predictive anomaly detection (AD) framework for solder joints. In surface mount technology (SMT), anomalous solder joints reduce the reliability of printed circuit boards (PCBs), which raises reworking expenses for PCB assembly lines. Therefore, predictive AD is essential to prevent solder joints with anomalies. The solder joint formation consists of three primary phases: solder paste printing, pick and place, and solder reflow. This research aims to predict the solder joint\textquotesingle s quality before the solder reflow phase by a novel framework, predictive AD for solder joints, which is called PADS. PADS first extracts 65 solder-joint-related features from datasets, then learns normal solder joints\textquotesingle patterns by reconstructing these features, and finally identifies a sample as an anomaly if its reconstruction error exceeds a designated threshold. The uniqueness of PADS is the utilization of novel features generated from interpreting the existing physics-based models and substantial real-world data acquired from SMT inspection machines, i.e., solder paste inspection (SPI) and automatic optical inspection (AOI). PADS has been extensively evaluated with commercial chip resistors R0402M (0.4 \texttimes 0.2 mm), R0603M (0.6 \texttimes 0.3 mm), and R1005M (1.0 \texttimes 0.5 mm), as well as SAC305 solder paste. The experimental results indicate that these novel features enable PADS to perform better in anomaly prediction for solder joints, and PADS outperforms many competitive baselines in prediction accuracy. \textcopyright 2011-2012 IEEE.

Abstract: The quality inspection of products before delivery plays a critical role in ensuring manufacturing quality. Quick and accurate inspection of samples is realized by highly automated inspection based on pattern recognition in smart manufacturing. Conventional ensemble methods have been demonstrated to be effective for defect detection. This study further proposed synthetic mechanisms based on using various features and learning classifiers. A database of 6000 sample images of printed circuit board (PCB) connectors collected from factories was compiled. A novel confidence synthesis mechanism was proposed to prescreen images using deep learning features. Spatially connected texture features were then used to reclassify images with low reliabilities. The synthetic mechanism was found to outperform a single classifier. In particular, the highest improvement in accuracy (from 96.00 to 97.83\%) was obtained using the confidence-based synthesis. The synthetic mechanism can be used to achieve high accuracy in defect detection and make automation in smart manufacturing more practicable. \textcopyright The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2024.

Abstract: This study aims to develop a scale-free classification-based defect diagnosis model for the pick-and-place (P\&P) machine in Surface Mount Technology (SMT) assembly line. SMT is a manufacturing process used to assemble Printed Circuit Boards (PCBs). The P\&P process is the primary procedure that follows the application of solder paste or adhesive to the board. Generally, the industry uses an automatic optical inspection (AOI) machine to detect assembly defects just after the P\&P process. However, inspection data from the AOI machine can only identify assembly defects; it cannot reveal the underlying causes of assembly failure. By conducting experiments with initial machine defects, it is possible to identify patterns associated with various root causes. Using the AOI and machine performance data, it is possible to trace the root causes of assembly defects using various machine learning methods. As the number of components used in the SMT assembly line increases, processing design of experiments (DOE), collecting sufficient data, and developing a defect diagnosis model for each type of component becomes time-consuming. The proposed model is trained on a single component type and then applied to other component types. Using the proposed model, when a new component is applied, the identification accuracies are more than 75.00\% for most of the root causes without conducting DOE. It can significantly reduce the time required to process experiments, collect data, and adjust models for new types of components. \textcopyright 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Abstract: We developed a visual inspection method for PCB board using an anomaly detection model. To improve feature extraction performance, we developed and optimized the feature extractor by comparing three types of backbone models. Then we compared two anomaly detection models with developed feature extractor as a backbone for visual inspection. Finally, we found the optimized loss function named mean-shifted contrastive loss which showed the highest accuracy in our experiment. Copyright \textcopyright The Korean Institute of Electrical Engineers.

Abstract: Printed circuit boards (PCB) defect detection is very important in today\textquotesingle s developed manufacturing industry, the surface defects of a PCB not only affect the aesthetics and comfort of the product, but generally also have adverse effects on its usability. Traditional inspection methods rely on the intuitive visual judgement of the operator, with strong subjective judgement factors, low accuracy rates and high staff costs. Deep learning can be a good solution to this problem to reduce the investment in human and material resources. A method/network structure for PCB defect detection based on deformable convolution has been proposed. The aim is to address the problems associated with existing PCB defect detection algorithms, which include low detection accuracy, poor real-time performance, and insufficient robustness. The network depth is increased to extract more and higher-level semantic information, which is guaranteed to be gradient-free under the resnet feature improved. The improved network model has demonstrated enhanced performance in the field of PCB defect detection. Specifically, it has outperformed the original RetinaNet network model by achieving a 3\% increase in mean accuracy(mAP) on the PCB defect detection dataset. Additionally, the detection speed of the modified model satisfies real-time requirements, further demonstrating its practical utility. \textcopyright 2023 IEEE.

Abstract: Testing and visual inspection of printed circuit boards (PCBs) and printed circuit board assemblies (PCBAs) are important procedures in the manufacturing process of electronic modules and devices related to locating and identifying possible defects and failures. Earlier defects detection leads to decreasing expenses, time and used resources to produce high quality electronics. In this paper an exploration and analysis about the current research regarding methods for PCB and PCBA testing, techniques for defects detection and vusial inspection is performed. The impact of machine and deep learning for testing and visual inspection procedures is also investigated. The used methodology comprises bibliometric approach and content analysis of papers, indexed in scientific database Scopus, considering the queries: \textquotedblleft PCB and testing\textquotedblright and \textquotedblleft PCB and testing\textquotedblright , \textquotedblleft printed circuit board assembly and testing\textquotedblright and \textquotedblleft PCBA and testing\textquotedblright , \textquotedblleft PCB defect detection\textquotedblright and \textquotedblleft PCBA defect detection\textquotedblright , \textquotedblleft PCB and visual inspection\textquotedblright , and \textquotedblleft PCBA and visual inspection\textquotedblright . The findings are presented in the form of a framework, which summarizes the contemporary landscape of methods for PCBs and PCBAs testing and visual inspection. \textcopyright 2024, Institute of Advanced Engineering and Science. All rights reserved.

Abstract: Many devices, including digital cameras, washing machines, smart cars, computers, etc., will have a PCB as an integral component. Today’s PCB demands the highest density of components and increased tolerance on these components. Surface mount technology (SMT) is making it possible to add many miniaturized components to PCBs. Component size is decreasing, and the component number is increasing; inspecting these components is becoming even more challenging. As many products are becoming smart, there is a huge demand for PCBs. Huge demands for PCBs also demand quicker inspection techniques. This shows the increased demand for online and inspection techniques. Many types of defects are encountered in the manufacturing of PCBs, such as missing components, cracks, etc. Many of the existing inspection techniques are not suitable for online PCB evaluation. In this scenario, the method presented in this article becomes very important. The present method uses image processing and deep learning-based technique for image identification and classification. Also, the method helps in the real-time assessment of PCBs. \textcopyright 2024 American Institute of Physics Inc.. All rights reserved.

Abstract: In electronics manufacturing, solder joint defects are a common problem affecting a variety of printed circuit board components. To identify and correct solder joint defects, the solder joints on a circuit board are typically inspected manually by trained human inspectors, which is a very time-consuming and error-prone process. To improve both inspection efficiency and accuracy, in this work we describe an explainable deep learning-based visual quality inspection system tailored for visual inspection of solder joints in electronics manufacturing environments. At the core of this system is an explainable solder joint defect identification system called SolderNet which we design and implement with trust and transparency in mind. While several challenges remain before the full system can be developed and deployed, this study presents important progress towards trustworthy visual inspection of solder joints in electronics manufacturing. Copyright \textcopyright 2023, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved.

Abstract: The printed circuit board (PCB) plays a critical role in any electronic product, and its manufacturing quality assurance is responsible for substantially impacting the final product’s price. Hence, research and development (R\&D) for better inspection methods have been an internationally growing subject. Regarding the multiple PCB inspection paradigms, the visual investigation of these boards represents the majority of modern techniques. Thus, the present survey collects the most impacting studies over the last 25 years, highlighting their operation strategies and how they evolved until recently, including introducing artificial intelligence (AI) approaches to increase their overall performance and lower operating costs. Furthermore, this work calls attention to the importance of public PCB datasets to contribute to developing such methods and provide benchmarking references. Finally, the most recent challenges in the field are described and discussed. \textcopyright The Author(s) 2024.

Abstract: Printed circuit board (PCB) manufacturing processes are becoming increasingly complex, where even minor defects can impair product performance and yield rates. Precisely identifying PCB defects is critical but remains challenging. Traditional PCB defect detection methods, such as visual inspection and automated technologies, have limitations. While defects can be readily identified based on symmetry, the operational aspect proves to be quite challenging. Deep learning has shown promise in defect detection; however, current deep learning models for PCB defect detection still face issues like large model size, slow detection speed, and suboptimal accuracy. This paper proposes a lightweight YOLOv8 (You Only Look Once version 8)-based model called LW-YOLO (Lightweight You Only Look Once) to address these limitations. Specifically, LW-YOLO incorporates a bidirectional feature pyramid network for multiscale feature fusion, a Partial Convolution module to reduce redundant calculations, and a Minimum Point Distance Intersection over Union loss function to simplify optimization and improve accuracy. Based on the experimental data, LW-YOLO achieved an mAP0.5 of 96.4\%, which is 2.2 percentage points higher than YOLOv8; the precision reached 97.1\%, surpassing YOLOv8 by 1.7 percentage points; and at the same time, LW-YOLO achieved an FPS of 141.5. The proposed strategies effectively enhance efficiency and accuracy for deep-learning-based PCB defect detection. \textcopyright 2024 by the authors.

Abstract: In recent years, with the research and development of deep learning, it has been more widely used in various fields, becoming an important productivity tool. In electronic manufacturing, an adaptive automatic optical inspection (AOI) system is proposed for defect detection of printed circuit board components (SMD-PCB), a key part of the industry chain. It is a combination of AOI based on traditional computer vision and multimodal imaging and PatchCore, a one-class novelty detection method based on deep learning. It aims to utilize one-class novelty detection method to detect and avoid defect missing due to different and variable objects and defects that require a lot of manpower to eliminate in normal AOI. Due to the unique characteristics of industrial applications, it is important to ensure inspection quality while controlling hardware and time costs. Therefore, in this adaptive AOI system, the reduction of time and hardware consumption of deep learning-based PatchCore during the training and inference process becomes an important part of putting it into practical application. In this work, we investigate the lightweighting of the PatchCore method, whose core idea is to use deep convolutional neural networks (CNN) for feature extraction to construct a feature memory bank, and then retrieve the feature memory bank to determine the novelty of the samples to be tested. According to this principle, we mainly lighten the PatchCore method in two directions: lightning the structure of the feature extractor and selecting the main features of feature memory bank, and then compare its performance with the original version. The experimental results show an order of magnitude reduction in hardware and time consumption reduction, while the performance remains almost the same. \textcopyright 2025

Abstract: Accurate defect detection in industrial automated optical inspection (AOI) is crucial as it directly affects product quality and production efficiency. Although numerous techniques have been developed for industrial defect detection, most of them rely on single-texture image data. This dependence limits the accuracy and robustness of the defect detection due to inadequate optical source information. To overcome the problem of low accuracy owing to the lack of 3D topographic information, a multidimensional information fusion (MIF) module is proposed that fuses texture image and depth image features. The MIF module includes tailored mechanisms to fully extract complementary semantic information from space and channel dimensions. A hierarchical fusion strategy further improves feature integration by enabling higher-layer feature fusion via lower-layer Transformer blocks and efficiently removing redundant features. Afterward, feature extraction is performed on the fused feature map, and output is obtained. To enhance the detection accuracy, the position information mask (PIM) module is introduced for post-processing. The PIM module uses surface mount devices (SMDs) position data from Gerber files to create a position information mask. The mask helps filter out defects that are often misidentified owing to missing design information. The results of the comparative experiments demonstrate that the average accuracy of our method on the printed circuit board assembly (PCBA) defect dataset is 99.93\%, which is 5.64\% higher than that of conventional YOLOV5. Furthermore, a comprehensive ablation study is conducted to elucidate the contribution of the proposed MIF and PIM modules. It demonstrated that the present model serves as a valuable reference for PCBA surface defect detection. \textcopyright 2025 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement.

Abstract: In electronics manufacturing, solder joint defects are a common problem affecting a variety of printed circuit board components. To identify and correct solder joint defects, the solder joints on a circuit board are typically inspected manually by trained human inspectors, which is a very time-consuming and error-prone process. To improve both inspection efficiency and accuracy, in this work, we describe an explainable deep learning-based visual quality inspection system tailored for visual inspection of solder joints in electronics manufacturing environments. At the core of this system is an explainable solder joint defect identification system called SolderNet that we design and implement with trust and transparency in mind. While several challenges remain before the full system can be developed and deployed, this study presents important progress towards trustworthy visual inspection of solder joints in electronics manufacturing. \textcopyright 2023 The Authors. AI Magazine published by John Wiley \& Sons Ltd on behalf of Association for the Advancement of Artificial Intelligence.

Abstract: Printed circuit boards (PCBs) are primarily used to connect electronic components to each other. It is one of the most important stages in the manufacturing of electronic products. A small defect in the PCB can make the final product inoperable. Therefore, careful and meticulous defect detection steps are necessary and indispensable in the PCB manufacturing process. The detection methods can generally be divided into manual inspection and automatic optical inspection (AOI). The main disadvantage of manual detection is that the detection speed is too slow, resulting in a waste of human resources and costs. Thus, in order to speed up the production speed, AOI techniques have been adopted by many PCB manufacturers. Most current AOI mechanisms use traditional optical algorithms. These algorithms can easily lead to misjudgments due to different light and shadow changes caused by slight differences in PCB placement or solder amount so that qualified PCBs are judged as defective products, which is also the main reason for the high misjudgment rate of AOI detection. In order to effectively solve the problem of AOI misjudgment, manual re-judgment is currently the reinspection method adopted by most PCB manufacturers for defective products judged by AOI. Undoubtedly, the need for inspectors is another kind of labor cost. To reduce the labor cost. of manual re-judgement, an accurate and efficient PCB defect reinspection mechanism based on deep learning algorithm is proposed. This mechanism mainly establishes two detection models, which can classify the defects of the product. When both models have basic recognition capabilities, the two models are then combined into a main model to improve the accuracy of defect detection. In the study, the data provided by Lite-On Technology Co., Ltd. were implemented. To achieve the practical application value in the industry, this research not only considers the problem of detection accuracy, but also considers the problem of detection execution speed. Therefore, fewer parameters are used in the construction of the model. The research results show that the accuracy rate of defect detection is about 95\%, and the recall rate is 94\%. Compared with other detection modules, the execution speed is greatly improved. The detection time of each image is only 0.027 s, which fully meets the purpose of industrial practical application. \textcopyright 2023 by the authors.

Abstract: The essential requirements for any digital holography-based system to get employed as an inspection device in electronic industry are portability and ability to automatically focus on the reconstruction plane. Any deviation from the exact propagation distance between the hologram plane and the reconstruction plane may result in erroneous measurement of the surface features under inspection. We report for the first time, to the best of our knowledge, the autofocus capability in a digital holography-based portable camera for full-field quantitative inspection of solder joint defects in a printed circuit board (PCB). An intensity gradient-based algorithm is developed and applied to the digital hologram to automatically locate the reconstruction plane and determine the propagation distance. For a given range of distance, the algorithm calculates the maximum value of high-frequency components in the reconstructed image, and the corresponding propagation distance is used in the reconstruction. The autofocus algorithm along with the double exposure digital holographic interferometry (DHI) is applied on the PCB under thermal loading. 2-D and 3-D deformation maps are analyzed to detect different kinds of solder joint defects, such as holes, irregular shapes, excess soldering, and bridging. The work presented in this article may further stimulate the development of digital holographic techniques for several applications in metrology and inspection in the electronics manufacturing industry. \textcopyright 1963-2012 IEEE.

Abstract: Printed circuit boards (PCBs) are an important component of electronic devices. Therefore, ensuring the quality of such PCBs in the manufacturing process is crucial. Especially, cracks or scratches appearing on the PCB surface pose a significant hurdle, due to their minuscule size, making them the most challenging to address. In this work, we present a real-time automated algorithm for defects inspection of printed circuit boards (PCBs) in different lighting conditions. First, the Oriented FAST and Rotated BRIEF (ORB) algorithm extracts features from the input images, then the Brute-force matching method matches these features with the ORB features template. Next, the input images are calibrated to match the size and orientation of the template data by the RANSAC (Random Sample Consensus) algorithm. Finally, the defective areas on the PCB surface are segmented by using the U-NET (i.e., a type of convolutional neural network (CNN)) model. The proposed algorithm is tested in three different lighting conditions: low light, normal light, and high light conditions. Experimental studies are conducted on a representative PCB to evaluate the defect detection capacity of the proposed algorithm and the experimental results show that the proposed system works well in the three different lighting conditions with an accuracy of up to 97\%, the detection speed is 12 frames per second (FPS). \textcopyright 2025

Abstract: The ever-expanding market for electronic devices has significantly heightened the demand for high-quality printed circuit boards (PCBs). Even minor defects in PCBs can pose substantial safety risks for end-users. This article provides a comprehensive review on deep learning-based approaches for PCB defect detection. Our exploration covers various critical aspects, including the classification of PCB defects, automated vision inspection (AVI) techniques, object detection methodologies, and the widespread adoption of deep learning models. Specifically, we focus on the state-of-the-art approach known as region-based Fully Convolutional Network with Feature Pyramid Networks (FPN-RFCN). Additionally, we discuss effective data augmentation techniques and commonly used evaluation metrics in this domain. This review provides valuable insights for researchers, practitioners, and industry professionals engaged in PCB quality assurance. \textcopyright 2025, Malque Publishing. All rights reserved.

Abstract: The integration of Artificial Intelligence (AI) with Automated Optical Inspection (AOI) systems holds significant potential for improving the accuracy and efficiency of defect detection in Printed Circuit Board Assembly (PCBA). This paper presents the findings of an industry-wide study conducted by the International Electronics Manufacturing Initiative (iNEMI) on the adoption and enhancement of AI in AOI systems. Building on the results of Phase 1, which established performance metrics and identified key challenges, Phase 2 involved experimental evaluations to provide industry-wide recommendations. The study highlights the benefits, challenges, and future directions for AI-enhanced AOI systems. \textcopyright 2024 IEEE.

Abstract: In electronics manufacturing, the inspection of defects of electrical components on printed circuit boards (SMD-PCB) is an import part of the production chain. This process is normally implemented by automatic optical inspection (AOI) systems based on classical computer vision and multimodal imaging. Despite the highly developed image processing, misclassifications can occur due to the different, variable appearance of objects and defects and constantly emerging defect types, which can only be avoided by constant manual supervision and adaption. Therefore, a lot of manpower is needed to do this or to perform a subjective follow-up. In this paper, we present a new method using the principle of multimodal deep learning-based one-class novelty-detection to support AOIs and operators to detect defects more accurate or to determine whether something needs to be changed. By combining with a given AOI classification a powerful adaptive AOI system can be realized. To evaluate the performance of the multimodal novelty-detector, we conducted experiments with SMD-PCB-components imaged in texture and geometric modalities. Based on the idea of one-class-detection only normal data is needed to form training sets. Annotated defect data which is normally only insufficiently available, is only used in the tests. We report about some experiments in accordance with the consistence of data categories to investigate the applicability of this approach in different scenarios. Hereby we compared different state-of-the-art one-class novelty detection techniques using image data of different modalities. Besides the influence of different data fusion methods are discussed to find a good way to use this data and to show the benefits using multimodal data. Our experiments show an outstanding performance of defect detection using multimodal data based on our approach. Our best value of the widely known AUROC reaches more than 0.99 with real test data. \textcopyright 2023 SPIE.

Abstract: The printed circuit boards (PCBs) are crucial components of most electronic devices. In the last decades, the PCBs’ manufacturing process was significantly improved, mainly by surface mounted technology (SMT) and automatic optical inspection (AOI) implementation. The real data as an output from the AOI device used for our analysis have been composed in a real manufacturing company. The currently used AOI solution achieves an accuracy of 95.82\%. The goal of our study was to train an artificial neural network (ANN) to detect the defect PCBs with the highest possible accuracy. Different approaches have been used for ANN training, such as the experimental approach, regression, and Taguchi method. The resulted PCA-ANN model combines principal components analysis (PCA) method for data dimensionality reduction and ANN for low quality products detection. Our proposed model increases the AOI accuracy rate by 3.95\%. Copyright \textcopyright 2025 Inderscience Enterprises Ltd.

Abstract: With the rapid growth of the PCB manufacturing industry, there is an increasing demand for computer vision inspection to detect defects during production. Improving the accuracy and generalization of PCB defect detection models remains a significant challenge. This paper proposes a high-precision, robust, and real-time end-to-end method for PCB defect detection based on deep Convolutional Neural Networks (CNN). Traditional methods often suffer from low accuracy and limited applicability. We propose a novel approach combining YOLOv5 and multiscale modules for hierarchical residual-like connections. In PCB defect detection, noise can confuse the background and small targets. The YOLOv5 model provides a strong foundation with its real-time processing and accurate object detection capabilities. The multi-scale module extends traditional approaches by incorporating hierarchical residual-like connections within a single block, enabling multiscale feature extraction. This plug-and-play module significantly enhances performance by extracting features at multiple scales and levels, which are useful for identifying defects of varying sizes and complexities. Our multi-scale architecture integrates feature extraction, defect localization, and classification into a unified network. Experiments on a large-scale PCB dataset demonstrate significant improvements in precision, recall, and F1-score compared to existing methods. This work advances computer vision inspection for PCB defect detection, providing a reliable solution for high-precision, robust, real-time, and domain-adaptive defect detection in the PCB manufacturing industry. \textcopyright The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

Abstract: Defect detection is a critical element in the PCB manufacturing process. Different from surface mount PCB, the components on the plug-in PCB are usually installed manually, resulting in significant errors. We make contributions in the following two aspects: (1) a framework and measurement method of a light source and make a cheap and efficient lighting system; (2) a fusion algorithm based on machine learning and morphology for polarity detection of plug-in capacitors. The capacitor is detected using SVM and fused with the polar coordinate expansion method. The AOI system and the proposed fusion algorithm have been applied to the production line, with an accuracy of 99.73 \% and a missed detection rate 0.12 \%. \textcopyright 2023, The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature.

Abstract: Traditional approaches to defect detection, relying on rule-based algorithms or human visual examination, are inadequate and ineffectual in handling intricate fault patterns. Implementing computer vision techniques in PCB defect detection can improve inspection processes\textquotesingle accuracy, reliability, and effectiveness. Deep learning is now used in those processes to explore important technology. This study presents a novel automated method for defect identification using deep learning methods. The system is built utilizing AlexNet, U-Net, ResNet 101, and Inception-v3, prominent frameworks known for their outstanding capacity to extract pertinent visual data components. The proposed technique comprehensively composes PCB photos, including faulty and defect-free samples. This training enhances the system\textquotesingle s capacity to precisely differentiate between conventional PCB designs and six additional categories of errors. The suggested architecture of the CNN models consists of double process layers represented as ConV I, Pooling I, ConV II, and Pooling II. The validation procedure is carried out by using a range of learning rate data. The implementation findings indicate that the ResNet-101 variant achieves the highest testing accuracy of 99.286\%. Indeed, incorporating this technology into the PCB manufacturing process enables the prompt identification of defective PCBs, leading to less waste, enhanced product quality, and heightened satisfaction among end-users. \textcopyright 2025 IEEE.

Abstract: The study aims to develop an automated system leveraging modern machine learning techniques to identify defects in printed circuit boards (PCBs). Defects such as shorts, spurious copper, and missing holes can significantly compromise electronic devices’ reliability and performance. Traditional manual inspection methods are time-consuming and error prone. Thus, this work addresses these issues by using a custom PCB defect dataset annotated with Roboflow and training a YOLOv5 model on Google Colab. The system integrates an Arducam Camera Module IMX219 with a Raspberry Pi 4 to capture high-resolution PCB images, which are then inspected in real-time for defects. By combining advanced object detection algorithms with affordable hardware, this method offers a practical and cost-effective solution for PCB defect detection. Results demonstrated that the system is highly accurate and efficient in detecting PCB flaws, with performance metrics of 0.92, 0.89, and 0.90 for precision, recall, and F1 score, respectively. This significantly enhances quality control in PCB manufacturing. \textcopyright The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

Abstract: We present a system focused on the Visual Inspection of Pin Through Hole (PTH) electronic components. The project was developed in a partnership with a multinational Printed Circuit Board Printed Circuit Board (PCB) manufacturing company which requested a solution capable of operating adequately on unseen components, not included in the initial image database used for model training. Traditionally, visual inspection was mostly performed with pre-determined feature engineering which is inadequate for a flexible solution. Hence, we used a one-shot-learning approach based on Siamese Neural Network model trained on anchor-negative-positive triplets. Using a specifically designed web crawler we collected a new and comprehensive database composed of electronic components which is used in extensive experiments for hyperparameters tuning on training and validations stages, achieving satisfactory performance. A web application is also presented, which is responsible for the management of operators, recipes, part number, etc. A hardware responsible for attaching the PCBs and a 4K camera is also developed and deployed on industrial environment. The overall system is deployed in a PCB manufacturing plant and its functionality is demonstrated in a relevant scenario, reaching a level 6 in Technology Readiness Level (TRL). \textcopyright 2023 by SCITEPRESS - Science and Technology Publications, Lda.

Abstract: The surface defects of printed circuit boards (PCBs) generated during the manufacturing process have an adverse effect on product quality, which further directly affects the stability and reliability of equipment performance. However, there are still great challenges in accurately recognizing tiny defects on the surface of PCB under the complex background due to its compact layout. To address the problem, a novel YOLO-HorNet-MCBAM-CARAFE (YOLO-HMC) network based on improved YOLOv5 framework is proposed in this article to identify the tiny-size PCB defect more accurately and efficiently with fewer model parameters. First, the backbone part adopts the HorNet for enhancing the feature extraction ability and deepening the information interaction. Second, an improved multiple convolutional block attention module (MCBAM) is designed to improve the ability of the model to highlight the defect location from a highly similar PCB substrate background. Third, the content-aware reassembly of features (CARAFE) is used to replace the up-sampling layer for fully aggregating the contextual semantic information of PCB images in a large receptive field. Moreover, aiming at the difference between PCB defect detection and natural detection, the original model detection head (DH) is optimized to ensure that YOLOv5 can accurately detect PCB tiny defects. Extensive experiments on PCB defect public datasets have demonstrated a significant advantage compared with several state-of-the-art models, whose mean average precision (mAP) can reach 98.6\%, verifying the accuracy and applicability of the proposed YOLO-HMC. \textcopyright 1963-2012 IEEE.

Abstract: Ensuring the quality and functionality of printed circuit boards (PCBs) during manufacturing requires precise, automated visual inspection. Detecting integrated circuits (ICs) on PCBs poses a significant challenge due to diverse component sizes, types, and intricate board markings that complicate accurate object detection. This study addresses this challenge by proposing an enhanced EfficientNet-YOLOv4 algorithm tailored explicitly for the IC detection of PCBs. Numerous modifications are integrated into YOLOv4, with the replacement of its original backbone by a robust feature extraction network, EfficientNetv2-L, and meticulous hyperparameter tuning, including variations in loss functions, anchor size configurations, and other training techniques. The methodology further incorporates diverse data augmentation techniques to enrich the training dataset and enhance the model\textquotesingle s generalization ability. Extensive experiments conducted in this study showed the efficacy and robustness of the algorithm in handling complex PCB layouts and varying lighting conditions, outperforming existing PCB inspection models. The proposed method, EfficientNetv2-L-YOLOv4, achieved an impressive F1-score of 99.22 with an inference speed of 0.14 s per image. The proposed method also performed well compared to EfficientNet-B7-FasterRCNN and the original YOLOv4; it attains an F1-score of 98.96 and an inference speed of 0.10 s per image (with a batch size of 4). These results highlight the significance of effective feature extraction networks for object detection. Beyond addressing IC detection challenges, this algorithm advances the fields of computer vision and object detection. The implementation of EfficientNetv2-L-YOLOv4 in real manufacturing scenarios holds promise for automating component inspections and potentially eliminating the need for human intervention. \textcopyright 2013 IEEE.

Abstract: Inspection in the manufacturing industry is commonly used to detect any abnormality or non-conformance in entire processes. Human visual inspection where the operator conducts a 100\% inspection of the product under a magnifying lamp. This kind of inspection affects the quality of the product due operator’s inconsistent product judgment which highly depends on experience and skills. In this paper, Automated Optical Inspection (AOI) image capturing will introduce to improve the monthly production output. MATLAB machine vision algorithms such as image masking and segmentation will process the captured image. Image boundaries will calculate the solder area to determine if the product met the defined specification. 95\% accuracy result on cross and hold-out validation using K Nearest Neighbor (KNN) supervised learning classifier algorithm. 30\textendash 40\% monthly productivity improvement after implementation of the AOI machine vision system. \textcopyright 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Abstract: While the automated optical inspection (AOI) was adopted for the surface mounted technology (SMT) on printing circuit board assembly (PCBA), it inspected that the products with miniaturized components and high-density placement of components on boards are becoming important issues. Especially, due to the manufacturing trend in the high-mix and low volume, the detection of the component marks with wide variation are growing into even more challengeable. As for the defect detection of AOI machine by using the similarity technology, it is very sensitive to different marks, including the different brightness of the font, spacing, brand, word shift and direction. Therefore, it implies that the AOI performance which is reduced as the AOI engineer fails to solve the demanding problem, and results in more time on checking the alarmed images by a human operator. Also, it affects the production yield rate and overall equipment effectiveness. Consequently, in this paper, a novel architecture is proposed to solve such a problem. That is, the classification method combined with similarity measurement on basis of the deep learning structure. The technique we used takes the MobileNet lightweight model to modify as the output with double stages to prevent the escape from occurring. The proposed method is approved by the manufacturing line to produce the products due to its no escape occurred, easier model training and faster implement. The proposed method provides a high improved rate of the average 90\% with the zero-leakage rate in SMT process according to the experimental results. \textcopyright 2023 IEEE.

Abstract: A printed circuit board (PCB) functions as a substrate essential for interconnecting and securing electronic components. Its widespread integration is evident in modern electronic devices, spanning computers, cell phones, televisions, digital cameras, and diverse apparatus. Ensuring product quality mandates meticulous defect inspection, a task exacerbated by the heightened precision of contemporary circuit boards, intensifying the challenge of defect detection. Conventional algorithms, hampered by inefficiency and limited accuracy, fall short of usage benchmarks. In contrast, PCB defect detection algorithms rooted in deep learning hold promise for achieving heightened accuracy and efficiency, bolstered by their adeptness at discerning novel defect types. This review presents a comprehensive analysis of machine vision-based PCB defect detection algorithms, traversing the realms of machine learning and deep learning. It commences by contextualizing and elucidating the significance of such algorithms, followed by an extensive exploration of their evolution within the machine vision framework, encompassing classification, comparison, and analysis of algorithmic principles, strengths, and weaknesses. Moreover, the introduction of widely used PCB defect detection datasets and assessment indices enhances the evaluation of algorithmic performance. Currently, the detection accuracy can exceed 95\% at an Intersection over Union (IoU) of 0.5. Lastly, potential future research directions are identified to address the existing issues in the current algorithm. These directions include utilizing Transformers as a foundational framework for creating new algorithms and employing techniques like Generative Adversarial Networks (GANs) and reinforcement learning to enhance PCB defect detection performance. \textcopyright 2013 IEEE.

Abstract: Surface mount technology (SMT) is a procedure for mounting electronic components to the surface of printed circuit boards (PCBs). Although the SMT procedure is more reliable than the conventional through-hole mounting, many errors may occur in SMT lines. In this article, we surveyed methods presented for the optical inspection of solder joints on PCBs. The methods are grouped by problem-solving approach: reference-based, machine learning/computer vision, deep learning, and 3-D reconstruction. We compare and discuss these approaches according to their advantages and disadvantages, with a focus on the more recent deep learning and 3-D reconstruction techniques, and list the public datasets for solder joint inspection. Since defective samples rarely occur in the SMT manufacturing lines, public datasets have few defective samples that are essential for the inspection task. To fill this gap, we publish a public dataset with both normal and different types of SMT errors. Our dataset can be used as a benchmark to compare different algorithms. The dataset is available at https://github.com/furkanulger/solder-joint-dataset. \textcopyright 1963-2012 IEEE.

Abstract: To increase the reliability of the printed circuit board (PCB) manufacturing process, automated optical inspection is often employed for soldering defect detection. However, traditional approaches built on handcrafted features, predefined rules, or thresholds are often susceptible to the variation of the acquired images’ quality and give unstable performances. To solve this problem, a deep learning-based soldering defect detection method is developed in this article. Like many real-life deep learning applications, the number of available training samples is often limited. This creates a challenging low-data scenario, as deep learning typically requires massive data to perform well. To address this issue, a multitask learning model is proposed, namely, PCBMTL, that can simultaneously learn the classification and segmentation tasks under low-data regimes. By acquiring the segmentation knowledge, classification performance is substantially improved with few samples. To facilitate the study, a soldering defect image dataset, namely, PCBSPDefect, is built. It focuses on the dual in-line packages (DIP) at the PCB back side, DIP at the PCB front side, and flat flexible cables. Experimental results show that the proposed PCBMTL outperforms the best existing approaches by over 5\textendash 17\% of average accuracy for different datasets. \textcopyright 2023 The Authors. Advanced Intelligent Systems published by Wiley-VCH GmbH.

Abstract: The quality of a printed circuit board (PCB) is paramount towards ensuring proper functionality of electronic products. To achieve the required quality standards, substantial research and development efforts were invested to automate PCB inspection for defect detection, primarily using computer vision techniques. Despite these advancements, the accuracy of such techniques is often susceptible towards varying board and component size. Efforts to increase its accuracy especially for small or tiny defects on a PCB often lead to a tradeoff with reduced real-time performance, which in turn limits its applicability in the manufacturing industry. Hence, this paper puts forward an enhanced deep learning network which addresses the difficulty in inferring tiny or varying defects on a PCB in real-time. Our proposed enhancements consist of i) A novel multi-scale feature pyramid network to enhance tiny defect detection through context information inclusion; and ii) A refined complete intersection over union loss function to precisely encapsulate tiny defects. Experimental results on a publicly available PCB defects dataset demonstrate that our model achieves 99.17\% mean-average precision, while maintaining real-time inferencing speed at 90 frames per second. In addition, we introduce three trend detection algorithms which alert an operator when abnormal development of defect characteristics is detected. Each algorithm is responsible for localizing defect buildups, increasing defect size and increasing defect occurrences, respectively. As a whole, the proposed model is capable of performing accurate and reliable real-time PCB inspection with the aid of an automated alert capability. The dataset and trained models are available at: https://github.com/JiaLim98/YOLO-PCB. \textcopyright 2023 The Author(s)

Abstract: Almost every electronic product used regularly contains printed circuit boards, which in addition to being used for business purposes are also used for security applications. Manual visual inspection of anomalies and faults in circuit boards during manufacture and usage is extremely challenging. Due to a shortage of training data and the uncertainty of new abnormalities, identifying undiscovered flaws continues to be complicated. The YOLO-v5 technique on a customized PCB dataset is used in the study to incorporate computer vision to detect six potential PCB defects. The algorithm is designed to be feasible, deliver precise findings, and operate at a considerable pace to be effective. A technique of image subtraction is also implemented to detect flaws in printed circuit boards. The structural similarity index, a perception-based method, gauges how similar non-defective and defective PCB images are to one another. \textcopyright 2023 IEEE.

Abstract: Today, the process of producing a printed circuit board assembly (PCBA) is growing rapidly, and this process requires cutting-edge debugging and testing of the boards. The Automatic Optical Inspection (AOI) process detects defects in the boards, components, or solder pads using image processing and machine learning (ML) algorithms. Although state-of-the-art approaches for identifying defects are well developed, due to three main issues, the ML algorithms and datasets are incapable of fully integrating into industrial plants. These issues are privacy limitations for sharing data, the distribution shifts in the PCBA industry, and the absence of a degree of freedom for reproducible and modifiable synthetic datasets. This paper addresses these challenges and introduces \textquotedblleft ReProInspect\textquotedblright , a comprehensive framework designed to meet these requirements. ReProInspect uses fabrication files from the designed PCBs in the manufacturing line to automatically generate 3D models of the PCBAs. By incorporating various techniques, the framework introduces controlled defects into the PCBA, thereby creating reproducible and differentiable defect datasets. The quality data produced by this framework enables an improved detection and classification scenario for AOI in industrial applications. The initial results of ReProInspect are demonstrated and discussed through detailed instances. Finally, the paper also highlights future work to improve the current state of the framework. \textcopyright 2024, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Abstract: Printed circuit board (PCB) is a crucial component of advanced electronic devices, and its quality control cannot be overlooked. This paper presents a comprehensive review of machine vision-based surface defect detection methods for PCBs, addressing the transition from traditional image processing to advanced deep learning techniques. With the increasing complexity of PCB designs and the demand for high-precision manufacturing, automated defect detection has become critical for quality control. The study introduces nine public datasets for PCB surface inspection and fourteen common types of PCB surface defects, and provides an overview of commonly used performance evaluation metrics in the field of PCB defect detection. This paper systematically analyzes three categories of detection approaches: image processing-based methods, machine learning-based classifiers, and deep learning architectures. Furthermore, the review provides a comparative evaluation of representative works for each method, revealing that deep learning-based methods achieve state-of-the-art performance, while image processing methods and traditional machine learning approaches remain valuable in resource-constrained scenarios. We believe that the Transformer architecture has transformative potential in handling global defect patterns, while emphasizing the need for lightweight model architectures through techniques like knowledge distillation. This paper also mentions hybrid methods and ensemble approaches, which leverage the advantages of different detection techniques to achieve defect detection, improve detection accuracy, and reduce the probabilities of false positives and false negatives. Key challenges such as class imbalance, computational complexity, and detection of micro-defects are discussed, along with future directions including self-supervised learning, multi-modal learning, 3D defect detection, and edge-cloud collaborative systems. \textcopyright 2025

Abstract: Printed circuit board (PCB) inspection is as challenging issue in electronic manufacturing. Visual inspection is broadly used to perform this task and has been instigated by many researchers. This paper focuses on review of 2D vision-based methods in PCB inspection. Traditional and modern advances methods are explored, and their advantages and disadvantages are addressed. Moreover, current research gap through literature review investigation in this domain is addressed and potential solutions are described. Finally, direction for future studies is presented for further improvement of existing methods. \textcopyright The Author(s), under exclusive licence to The Optical Society of India 2023.

Abstract: Reliability and lifetime of specific electronics boards depends on the quality of manufacturing process. Especially soldering splashes in some areas of PCB (printed circuit board) can cause change of selected electrical parameters. Nowadays, the manual inspection is massively replaced by specialized visual systems checking the presence of different defects. The research carried out in this paper can be considered as industrial (industry requested) application of machine learning in automated object detection. Object of interest\textemdash solder splash\textemdash is characterized by its small area and similar properties (texture, color) as its surroundings. The aim of our research was to apply state-of-the art algorithms based on deep neural networks for detection such objects in relatively complex electronic board. The research compared seven different object detection models based on you-look-only-once (YOLO) and faster region based convolutional neural network architectures. Results show that our custom trained YOLOv8n detection model with 1.9 million parameters can detect solder splashes with low detection speed 90 ms and 96.6\% mean average precision. Based on these results, the use of deep neural networks can be useful for early detection of solder splashes and potentially lead to higher productivity and cost savings. \textcopyright 2023, The Author(s).

Abstract: Structural adhesives or fixing glues are typically applied to larger components on printed circuit boards (PCBs) to increase mechanical stability and minimize damage from vibration. Existing work tends to focus on component placement verification and solder joint analysis, etc. However, the detection of structural adhesives remains largely unexplored. This paper proposes a vision-based method for detecting structural adhesive defects on PCBs. The method uses HSV color segmentation to extract PCB regions, followed by Hough-transform-based morphological analysis to identify board features. The perspective transformation then extracts and rectifies the adhesive regions, and constructs an adhesive region template by detecting the standard adhesive area ratio in its corresponding adhesive region. Finally, template matching is used to detect the structural adhesives. The experimental results show that this approach can accurately detect the adhesive state of PCBs and identify the qualified/unqualified locations, providing an effective vision-based detection scheme for PCB manufacturing. The main contributions of this paper are as follows: (1) A vision-based structural adhesive detection method is proposed, and its detailed algorithm is presented. (2) The developed system includes a user-friendly visualization interface, streamlining the inspection workflow. (3) Actual experiments are performed to evaluate this study, and the results validate its effectiveness. \textcopyright 2025 by the authors.

Abstract: In traditional methods, solder paste inspection (SPI) of chips can be performed manually by trained operators using specialized microscopes or visual inspection systems. However, manual inspection is time-consuming, prone to human error, and is not suitable for high-volume or complex printed circuit board (PCB) designs. Therefore, we studied a 3C-YOLOv10n algorithm based on YOLOv10, which can significantly improve the detection effect while ensuring lightweight. First, replace the Bottleneck structure of the C2f (Cross Stage Feature Fusion) module with KAN (Kolmogorov-Arnold Networks), use KANConv combined with C2f to replace the original C2f module, and design a C2f-KAN module to solve the problem of the largest The problem of partial global feature information loss caused by pooling is improved, thereby improving the accuracy and interpretability of the model; secondly, the original PSA (Partial Self-Attention (partial self-attention) is replaced with the CoordAttention attention mechanism to better capture local and global relationships in space, improving the accuracy of the network with almost no additional computational overhead; finally, CARAFE is introduced The dynamic upsampling module improves the feature fusion ability of the neck network from the perspective of point sampling. A comparative analysis conducted on the PCB-AoI dataset revealed that the enhanced model exhibits reduced computational complexity in comparison to the original YOLOv10n. Furthermore, it demonstrated improvements in three performance metrics: Precision, Recall, and mAP@0.5, which increased by 3.8\%, 1.4\%, and 2.9\%, respectively. These findings substantiate the efficacy of the proposed methodology. The improved algorithm can accurately identify defects on abnormal circuit boards, ensuring that the solder paste is applied in the correct amount, correct position and quality, while reducing missed detections and false detections, thereby achieving a reliable and high-yield soldering process .

Abstract: With the rapid advancement of industrial automation technology, the quality of PCBs directly affects the performance of electronic products. To solve the problem of missed or repeated detections caused by unreasonable anchor frame design in PCB surface defect detection tasks based on anchor frame detection algorithms, this paper proposes a BBW-YOLOv8s based anchor frame free PCB surface defect detection model using the one-stage anchor frame free algorithm YOLOv8s as the benchmark network. The model first adds a Biformer attention mechanism with good performance for small object detection to the SPFF module and Neck layer of the YOLOv8s network, reducing the model\textquotesingle s attention to irrelevant information in PCB images and enhancing its ability to learn key features; Secondly, referring to the multi-scale feature fusion network Bi FPN, the output of shallow feature maps in the backbone network is connected across scales with the detection head, so that the detailed information of shallow feature maps is guided by the semantic information of deep feature maps, and the model\textquotesingle s understanding ability of PCB image content is improved; Finally, the cross entropy loss function in the original YOLOV8 model was replaced with Focal Loss to make the model more focused on difficult to classify targets, thereby improving overall accuracy.

Abstract: Printed circuit boards are an important part of electronic products. Ensuring the quality of printed circuit boards is an important task. Traditional target detection methods require cropping of the image, which leads to poor real-time detection and poor accuracy. To address these issues, we propose an improved defect detection method for YOLOv5s. We use the single-stage target detection network model YOLOv5s. Firstly, we design a set of small anchor frames with small target sizes. Second, we add a small target detection head to the network and fuse strong semantic information from the second layer of the network feature map. Third, we embed the CBAM module and the SPP module into the Neck layer of the network. the CBAM module helps the network to find regions of interest in the images. the SPP module fuses the multi-scale features of the targets. The purpose of the above three components is to solve the problem that small target features are easily lost. The experimental results show that the improved network has a precisi rate of 97.1\%, a recall rate of 89.1\%, an mAP0.5 of 94.5\% and an mAP0.5:0.95 of 48.5\%, all four metrics are higher than the YOLOv5s base model.

Abstract: As an essential component of modern machines, printed circuit board (PCB) is widely used in various electronic products. Its quality significantly affects the quality of products. However, the production process of PCB is often accompanied with defects. In this paper, a defect detection algorithm is proposed. Data augmentation such as flipping, shifting, brightness adjustment, rotation, and Guass noise are carried out to diversify the dataset. You only look once (YOLO) v5s is then introduced to detect the PCB defects. Through parameter tuning and optimization, a trained detection model is achieved. F1-score and mean average precision (mAP) are used to assess the performance of the model. The experiment results show that the mAP and F1-score are 99.3\% and 99.0\%, respectively. The model developed based on YOLO-v5s algorithm can achieve superior performance, which is competent to detect the defects of PCBs.

Abstract:

Abstract: With the improvement of integrated circuit integration, the industrial complexity of printed circuit boards is gradually increasing, and the defects generated in the manufacturing process are also more subtle and complex. Traditional target detection methods are difficult to meet the needs of industrial detection. In order to reduce the error detection and missed detection in the defect detection of printed circuit boards, and effectively improve the detection accuracy under the premise of ensuring the detection speed, a defect detection algorithm for printed circuit boards based on improved Yolov5s is proposed. Firstly, the BiFPN ( Bidirectional Feature Pyramid ) is combined in the neck network, and the PANet in the Yolov5s benchmark structure is replaced by the BIFPN structure, and the bidirectional cross-scale connection is used to enhance the feature representation ability. Secondly, the Transformer layer module is designed by using the attention mechanism, and the sparse Transformer is proposed. The deformable layer contains a multi-head attention structure, which reduces the time complexity of the self-attention mechanism operation. Finally, the SE ( Squeeze-and-Excitation ) attention module is embedded in the basic network architecture to improve the performance of the model, so that it can better focus on important features and improve the detection accuracy. The experimental results show that the average accuracy value ( mAP @ 0.5 ) of the improved Yolov5s model for defect detection of printed circuit boards reaches 97.7 \%, which is 2.3 \% higher than that before the model improvement, and can meet the requirements of industrial detection.

Abstract: In modern electronic devices, printed circuit boards play a crucial role as core components, and their quality is essential for device performance and reliability. Due to the complexity of PCB manufacturing processes, issues such as open circuits, short circuits, and solder defects often occur, affecting the overall quality of products. Traditional manual inspection methods are unable to meet high standards, leading to the gradual adoption of automated defect detection technology as the mainstream trend in the industry. This paper presents a network model for PCB defect detection, aimed at addressing the challenge of detecting small targets. We propose SmallObjectEnhancePyramid, which utilizes the P2 feature layer processed by SPDConv to obtain features rich in small object information and fuses them with P3. Then, we integrate features using CSP-OmniKernel to effectively learn feature representations from global to local. Additionally, we introduce RMTBlock to improve RepC3, reducing the computational burden of global information and model parameter quantity. This study not only improves the accuracy of PCB defect detection, but also helps optimize manufacturing processes and enhance production efficiency. Experimental results show that FPSPNet effectively improves the detection accuracy of printed circuit board defects, achieving an mAP0.5 value of 98.8\%, precision of 99.3\%, and recall of 97.3\% on the dataset. It also outperforms existing models in terms of performance.

Abstract: Printed Circuit Boards (PCBs) are the backbone of electronic devices, critical for the functionality and reliability of modern technological applications. Despite advancements in manufacturing techniques, the production of PCBs is susceptible to defects due to inherent technical limitations. These defects can significantly impact the performance and reliability of electronic products, making effective defect detection a paramount concern. This research introduces a novel approach to PCB defect detection and classification by employing advanced deep learning-based object detection networks. Focusing on two state-of-the-art target detection frameworks, this study evaluates their efficacy across two distinct PCB datasets, representing varied defect distributions. The experimental analysis demonstrates the superior accuracy of the proposed methods in identifying and classifying various types of PCB defects. By leveraging the robust capabilities of deep learning for object detection, this research offers significant contributions to the field of electronic manufacturing. This research presents a comprehensive evaluation of the detection networks, highlighting their strengths in enhancing the precision and efficiency of the PCB inspection process. The findings of this research provide valuable insights and advancements in addressing the critical challenge of defect detection in PCBs, thereby facilitating the improvement of electronic product quality and manufacturing yield. This work not only advances the knowledge base in PCB defect detection but also proposes a scalable and effective solution to optimize the inspection process in the electronic manufacturing industry.

Abstract: The printed circuit boards (PCBs) industry is one of the fastest-growing industries in recent decades. The PCB manufacturing process is highly complicated and severely affected by social factors, which makes it very important to conduct integrated inspection, assuring and improving the production quality. In this article, we propose an artificial systems, computational experiments, and parallel execution-based integrated inspection method in cyber\textendash physical\textendash social systems (CPSS) to realize smart manufacturing. In this inspection system, rather than simply performing modeling, analysis, and control, we perform descriptive intelligence to construct production processes with limited multimodal information, perform predictive intelligence to conduct defect detection and defect prediction, and perform prescriptive intelligence to achieve defect diagnosis and defect management. In this way, our inspection system could offer a learning and training platform for workers to master professional inspection skills, provide an experimentation and evaluation platform for product defect monitoring and early warnings, and supply guidance about defect management and control to improve manufacturing processes. For technical implementation, we leverage a Transformer-based foundation model to achieve knowledge reasoning and human\textendash computer interaction. As a result, we provide an innovative solution to cope with the challenges of quality inspection in current smart manufacturing, and expect its further applications in the PCB industry.

Abstract: With the rapid development of Industry 4.0 and intelligent manufacturing, product quality has become a key factor in the core competitiveness of enterprises. Especially in the field of electronics manufacturing, printed circuit board (PCB) as a core component, its quality directly affects the stability and reliability of the equipment. The traditional manual visual inspection method in PCB defect detection has limitations such as low efficiency and poor accuracy, which cannot meet the demand for efficient and high-precision detection in modern industrial production. In this paper, trap detection of PCB boards is based on the YOLO algorithm. The YOLOv9 algorithm achieves efficient target detection by integrating the Generalized Efficient Layer Aggregation Network (GELAN) and the Programmable Gradient Information (PGI), which improves the computational efficiency while retaining the key information. By comparing the detection effects of YOLOV9 and YOLOV8, it is shown that the YOLOv9 algorithm is able to achieve higher detection accuracy and realtime performance in PCB defect detection, which provides a reliable automated inspection solution for quality control in industrial production.

Abstract: This study develops an automated quality inspection system based on machine learning using the You Only Look Once version 8 (YOLOv8) object detection algorithm and deploys it on a Raspberry Pi for the quality inspection of Printed Circuit Boards (PCBs). PCBs are crucial components in modern electronics, such as smartphones, computers, and other devices. This research reports on testing the YOLOv8 algorithm for PCB quality inspection. Experiments applied YOLOv8 to a PCB testing dataset on a Raspberry Pi, demonstrating the system’s efficiency and reliability. The test results show the system\textquotesingle s ability to detect defects such as Pad Misalignment, Open Circuit, Solder Bridging, and Misplacement, with varying level of accuracy. Accuracy for Good Assembly (GA), Solder Bridging (SB), Open Circuit (OC), and Misplacement (MP) classes ranged from 70\% to 95\%, 73\% to 84\%, 67\% to 72\%, and 67\% to 98\%, respectively. These results indicate that this system effectively enhances the efficiency and accuracy of PCB quality inspection when implemented on a Raspberry Pi, potentially improving the overall quality of electronic products.

Abstract: Recently, there has been increasing academic interest in the problem of detecting surface defects on PCBs due to their significance in manufacturing processes. To address the issue of low detection accuracy in existing PCB surface defect detection models, this paper introduces a novel model called CC-YOLO, which enhances YOLOv7. The proposed model incorporates the CARAFE (Content-Aware ReAssembly of FEatures) operator as an up-sampling technique to minimize feature information loss in the input image and enhance the model’s feature extraction capability. Additionally, we propose the C2f-C module, which extends CSPDarknet53 to 2-Stage FPN-Cascade and strengthens it further by incorporating multiple Bottleneck modules to capture more comprehensive gradient flow information. We evaluate the performance of the enhanced YOLOv7-based model, CC-YOLO, on the publicly available PCB surface defect dataset from the Intelligent Robotics Laboratory at Peking University. The experimental results demonstrate that CC-YOLO achieves an accuracy of 9 8 . 9 \%, 2.3 \% higher than the original YOLOv7 model.

Abstract: Accurate detection and classification of mixed defects in printed circuit board (PCB) are critical for maintaining the quality and reliability of electronic product. This paper presents TR-YOLO, an enhanced version of the YOLOv5 model, which is developed for PCB defect inspection. TR-YOLO introduces a modified backbone structure incorporating a C3TR module, enabling effective extracting features at multiple scales while minimizing computational resources overhead. The findings from the experiments indicate that TR-YOLO attains a mean average precision (mAP) of 96.4\%, with 4.1 giga FLOPs and 1.80 million parameters. Compared to the YOLOv5 model, the proposed TR-YOLO improves mAP by 0.63\%, precision by 0.73\%, recall by 1.04\%, F1-score by 1.34\%, and overall accuracy by 0.17\%. Moreover, it reduces computational complexity by 2.38\% (i.e., reduced FLOPs) and enhances the speed by 7.61\%. These findings highlight TR-YOLO’s ability to deliver a balanced performance between detection accuracy and processing efficiency. The proposed model shows strong potential for real-time automatic optical inspection (AOI) in PCB manufacturing, contributing to improved product quality and production reliability.

Abstract: This paper presents the integration of several deep learning techniques for defect inspection of plated through-hole (PTH) on printed circuit boards (PCBs). In our proposed system, the object detection technology of You Only Look Once (YOLO) allocates the position of PTHs; a semi-automatic clustering mechanism distinguishes normal and defective PTHs for collecting training data. A convolution neural network based on the ResNet framework is established for detecting PTH defects. A retrain mechanism is designed for retracting misclassified PTHs and updating the recognition model. The generative adversarial network (GAN) is employed to deal with the insufficiency of negative training samples. The detection model can achieve 98.96\% accuracy with continuous retraining and data augmentation. Further, a filtering mechanism based on template analysis eliminates ambiguous outputs of GAN and improves the accuracy of the detection model. This study integrates these deep learning techniques to establish a PTH defect detection system tested in a PCB manufactory against real PTHs and gained an overall prediction accuracy of 99.48\%.

Abstract: With the rapid advancements in surface mount technology (SMT), microelectronic packaging products are becoming smaller, denser, and defect-free. This has led to stricter requirements for surface quality inspection. In the realm of industrial visual inspection, there is a growing demand for three-dimensional imaging technology to achieve high-precision and fast image acquisition, as well as for AI-based online inspection models to enhance real-time defect detection. This article primarily focuses on first- and second-level package products for SMT microelectronics and discusses the main optical imaging technologies, analyzing the principles and limitations. Next, this article analyzes the current research status of intelligent detection models based on deep learning in industrial environments and discusses current technical issues in data augmentation, small object learning, and model lightweight. The article also summarizes the characteristics and applications of the latest representative products, comparatively analyzes the gap between developed and developing countries. Finally, this article outlines four major challenges and future research trends in this field, aiming to promote the intelligent evolution of related visual inspection equipment.

Abstract: Identification of PCB board defects early in the manufacturing process is crucial, as PCB quality control plays an important role in the electronics manufacturing industry. Defective PCBs can lead to product failures, increased costs, and reliability issues. The idea proposes a machine learning program for the detection and classification of defects in PCBs using a YOLOv5 object detection algorithm and an Inception v3 CNNbased approach. The dataset consists of annotated PCB images for the YOLOv5 algorithm, and the images are categorized into six defect types: missing holes, mouse bite, open circuit, short circuit, spurs, and spurious copper. The preprocessed defect images are analyzed using the YOLOv5 model for defect location, while the Inception v3 CNN model performs precise defect classification. The results and the output clearly show that the proposed approach achieves high accuracy in both detection and classification tasks. Hence, the aim is to provide an efficient and automated solution for PCB inspection, outperforming conventional manual checks.